Skip to content

#编码

10 today

Oct 20Tuesday

Hacker News front page

Maxwell: A new kind of model that sits between code and hardware

doubleAI 推出代码执行模型 Maxwell,根据程序运行时的硬件与数据优化代码,并对每项改进进行形式化等价证明。它在 CPU 和 GPU 上运行,收集足够运行时数据后,在程序持续运行期间替换为针对当前机器的优化代码。页面称数据处理流水线加速 38 ×、AI 推理内核加速 4.25 ×,LTX-2.5 在两块 B200 上生成每段 4K 视频的耗时从 250s 降至 134 s。

Hacker News front page

Vite+ an excellent toolchain for AI agent development

Vite+ 在作者为期三周、覆盖四个项目的实践中,通过统一命令减少了 AI 智能体、开发者与 CI 之间的工作流歧义。作者用 vp create 创建项目,以 vp check、vp test 和 vp build 验收,并要求保留一个锁文件、避免重复引入格式化工具。

Computing Life · Share · Yage

When AI writes code faster than people read it, what's left for code review?

As AI speeds up coding, code review shifts toward checking business context, routing risk and keeping human accountability. A passing approval and green tests don't prove anyone understood the change. Anthropic splits review into pre-submit self-checks, multi-agent checks during the PR and risk tiers at merge, each needing 0, 1 or 2 human sign-offs, and model opinions can't replace a signature.

Why it matters: Anthropic splits code review into pre-submit self-checks, parallel defect finding and risk-tiered sign-off, pointing human attention at business assumptions and merge accountability.

Simon Willison

My obituary for programming

编码智能体让享受亲自解题的职业程序员失去部分工作乐趣,这是 Simon Willison 引用 Jeremy Scheff 文章时强调的核心观点。Scheff 回顾了程序员通过解决逻辑难题获得报酬,并自主制定策略、时间表和可行性边界的工作体验。Willison 指出,对重视构建可用系统的程序员,LLM 是持续带来收益的礼物;对把亲自解决问题视为重要乐趣的人,这种损失很真实。

Hacker News front page

How abliterated models can get you pwned

ProjectDiscovery 通过投毒开源模型并接入 Codex CLI,演示了修改后的权重如何在正常响应之外隐藏凭据外传后门。Qwen2.5-7B-Instruct 实验使用 500 条干净样本和 125 条投毒样本,触发成功率与干净请求准确率均为 100%(各 50/50),演示中外传的是 .env 内模拟生产环境的虚构凭据。

AI HOT · Products

Cursor launches Agent SDK for defining and deploying agents from a code folder

Cursor launched its Agent SDK, which lets teams define agents in a code folder, test them with evals, and deploy to Cursor or their own infrastructure.

Why it matters: Cursor's Agent SDK organizes prompts, tools and skills as a code project, testing agent behavior with local evals and replays of real pull requests.

Hacker News front page

Code is clay now

Alberto Denia 提出,智能体编码降低了修改代码的成本,让开发可以像塑形黏土一样先搭整体、再逐步打磨。工匠模式让人参与每轮迭代,适合探索想法和精修功能;工厂模式将判断提前嵌入规格、生成器与验证器组成的流水线,适合处理重复任务。两种模式可以混用,产品判断、架构和护栏仍不可省略。

Hacker News front page

The End of Rationing

编码智能体降低了软件构建成本,但软件产品仍沿用研发资源稀缺时期的设计与品类惯例,是 Sanket Saurav 的核心判断。他指出,使用 LLM 的产品承担持续的用户推理成本,软件经济结构正从高固定成本、低边际成本转向低固定成本、较高边际成本。他主张从基本原理重新思考代码协作与 CRM,并通过整合更多相关产品突破过去“只做一件事”的限制。

Hacker News front page

Ask HN: How to deal with shipping AI-written code you don't understand?

rzk 在 Ask HN 引用一位 Meta 工程师的困惑,询问如何应对交付自己并不充分理解的 AI 生成代码。被引用者担心逐行检查跟不上交付速度,只理解系统高层结构却不了解实现细节。评论中的应对方法包括人工定义架构后让 LLM 实现、团队共同审查设计及加强测试,也有人质疑测试能否替代代码理解与安全审查。

Hacker News front page

The reverts will continue until morale improves

PostgreSQL 19 进入 beta 期后回退了约 8 到 10 项重要功能,参与部分功能开发或提交的作者分析了可能原因。部分功能存在架构缺陷,LLM 辅助代码审查又加速发现大量边界问题;资深开发者忙于安全修复,也可能使常规审查推迟,社区最终选择维持发布计划。作者提出将 LLM 辅助模糊测试纳入功能开发、增强内部接口,并探索让实验性功能在主分支逐步成熟的机制。

TechCrunch · AI

OpenCode Console breached, leaked API keys rack up unauthorized AI charges

OpenCode says an intrusion into its legacy system hit 1.4% of OpenCode Console users, and leaked API keys led to unauthorized charges on AI accounts. Some users report bills above $1,000, and refunds are being processed. The open-source app and command-line tool were not affected, and affected users have been told to revoke and replace their keys.

Why it matters: The team's attack-code analysis was blocked by a model's safety filter, and switching to a model that refuses fewer defensive requests improved the investigation.

GitHub Blog · AI & ML

How to continue a GitHub Copilot app session in VS Code

GitHub Copilot app 会话可直接在 VS Code 中继续,保留原有任务、计划、对话、代码差异和工作树。教程说明如何通过 Open in VS Code 打开会话,逐行审查改动、继续与 Copilot 对话、手动编辑并运行本地测试。完成后在会话关联的分支上提交并推送,即可更新已有的 pull request,无需复制文件或手动重新关联。

Ars Technica · AI

AI game clones are making indie developers afraid to show their work

AI 生成的游戏仿制品让独立开发者陷入困境:不展示作品难以获得销量,公开开发进度又可能遭到抄袭。McPixel 开发者 Sos Sosowski 记录了多位同行的投诉,仿制 Holmér 旋转 Tetris 游戏的 Charlie Greenman 则认为自己的作品属于新创作。

Hacker News front page

Evidence (YC S21) Is Hiring Software Engineers

YC S21 孵化的商业智能公司 Evidence 正在招聘软件工程师,其开源项目已获 7,000 GitHub stars,客户覆盖初创公司到财富 100 强企业和政府机构,常替代 Tableau 和 Power BI。公司正推动由 AI 智能体承担大部分软件生命周期的开发流程,团队每周仅一次会议。

TechCrunch · AI

People are using AI to make their smart TVs dumb again

用户正借助 Claude Code 和 dumbtv、cleanyourtv 等项目改造智能电视,移除广告、精简预装软件并替换主屏幕。cleanyourtv 开发者称,该方案通过 Wi-Fi 操作,无需 root 权限且可恢复出厂状态;一名用户称禁用了 25 个冗余应用,将 192 MB 的原装启动器替换为 83 MB 的自定义启动器。

Hacker News front page

I no longer care to work on any Ruby gems so I've archived all my repositories

一名 Ruby gems 维护者宣布归档全部仓库、停止维护,原因是对 Ruby 生态转向 LLM 智能体失去热情。他批评 Rails 负责人使用智能体生成应用,以及 Matz 将 Ruby 定位为面向 AI 时代的编程语言。他认为,人类编写的开源代码正被大型企业用于训练模型,而模型生成的内容又污染了这一公共资源。

Oct 19Monday

AI HOT · Models

MiniMax releases M3.1 Flash with nearly 3x faster inference and 70% lower usage cost

MiniMax released M3.1 Flash, which it says runs online inference nearly three times faster and cuts overall usage cost on agent tasks by 70% versus the previous generation. The model scores 73.7% on DeepSWE v1.1 and reaches 200+ TPS in everyday output. Pricing is $0.15 per 1M input tokens, $0.60 per 1M output tokens and $0.015 per 1M cached input tokens.

Why it matters: MiniMax claims a 70% cut in agent-task usage cost over the prior generation, with benchmark results on complex tasks, coding and computer use at leading-model levels.

Hacker News front page

Superlite: A Faster Alternatie to Superpowers

Superlite 是作者开发的 Claude Code 插件,通过并行任务、分层智能体和代码钩子,将规格交给智能体后自动完成开发、测试与审查。当前版本单次基准用时 16.1 min,通过 177 / 177 项隐藏测试,相比 superpowers 原版此前 3 次运行的平均 33.6 min 减少 52% 时间;这一比较来自同机测试,并非多次重复的新版本结果。

Computing Life · Share · Yage

Anthropic's Scion source-control rebuild: which engineering calls transfer, which don't

Anthropic rebuilt its internal source control with Scion, and this analysis sorts the engineering calls other teams can reuse from the system design they shouldn't copy. The large monorepo's daily merges rose from 2,500 in May 2026 to more than 30,000 in September. Scion uses a thin client, server-side recomputation and three cache layers, and the report says source-control infrastructure CPU and memory use fell 200-fold.

Why it matters: The reusable parts for ordinary teams are pulling code on demand, reviewing the intent of a change and building automatic rollback guardrails.

Hacker News front page

I built a programming language that LLMs can't use

开发者构建并开源 KillSwitch,通过对抗性语言设计测试 LLM 和编码智能体能否理解、编写未见过的代码。它是一种基于栈的解释型语言,利用字符计数、规范中的隐藏谎言及触发安全护栏的语句增加难度,示例见 https://killswitch-lang.org 。

Hacker News front page

Sanity's Edge – A voice BBS from 1994

Sanity's Edge 将运行于 1994 年至 1997 年 3 月的电话语音 BBS 保存到网页,恢复了原始录音和交互体验。

Hacker News front page

Personal software factories are a waste of time

一位工程师公开劝退同行:不要花时间搭建个人“软件工厂”,除非是用于工作场景。他认为这类系统只是重复“立项、放弃、周一回去上班”的旧循环,不如直接打开聊天框让 AI 动手做产品;真正值得投入的是在工作中搭建自动化平台,因为只有那里才有深度的系统理解和真实复杂需求。

Hacker News front page

Someone behind the software

Linear 发文讨论软件产品为何越来越像流水线产物:两个团队面对同样的客户和技术,做出的东西却截然不同,差别来自创作者的经验与判断。作者认为智能体让个人独立产出改动变得更容易,但缺少统一视角后,产品会变成各项能力的拼凑,而非有性格的整体。

Computing Life · Share · Yage

相机、无人机与手机哪些已被破解,AI 在硬件逆向中做了什么

消费设备的解锁涉及官方开放、替代固件、隐藏参数、临时内存补丁和持久扩展五类机制,不能统称为攻破底层安全锁。适马 fp 的 fpSup 属于运行内存补丁,重启即失效;大疆 Mavic 4 Pro 的公开信号增强工具修改遥控器侧参数,未改飞机固件。

Oct 18Sunday

Hacker News front page

The Brain Sandwich

工程师 Emily 使用 Brain Sandwich 方法协作编码,流程是先自己理解问题,再交给 AI 执行,最后自己审查结果。启动 AI 智能体前先阅读并形成大致方向,以免被带向错误或无须解决的问题。提交他人审查前,应阅读 diff、判断复杂度是否合理并完善 PR 描述;如果无法解释改了什么以及为什么改,就还没准备好交给别人审查。

The Decoder

Anthropic cuts junior technical hiring as AI agents log six times the work hours of staff

Anthropic's internal research shows junior roles fell from 27% to 9% of new technical hires, which the company attributes to stronger AI agents. Junior roles in the technical division dropped from 21% to 14% since early 2025. By August 2026, agents logged six times the work hours of human staff, and 19 times in the research division.

Why it matters: Anthropic compares the technical division's shrinking junior hiring with near-stable non-technical hiring, with the cuts concentrated where Claude handles complex work.

The Decoder

Four Gemini 4 Argon agents beat Minecraft without any human help

Google DeepMind 团队的四个 Gemini 4 Argon 智能体在无人干预下击败末影龙,完成 Minecraft 通关。团队将它们置于随机生成的世界,未指定角色、策略或游戏专用工具;智能体自行分工、互相警告威胁、回收死亡队友的物品,并在共享记忆中记录失误。

Hacker News front page

Anthropic report: junior technical hiring share fell from 27% to 9% as AI agent use grew

Anthropic's technical report links wider AI agent use to a smaller share of junior technical hires, higher employee output and a shift toward managing agents. Drawing on internal data and 17 employee interviews, it found the junior technical hiring share fell from 27% to 9% since early 2025, and by August agent work led by employees ran six times longer than the employees' own.

Why it matters: Anthropic's internal study finds staff moving toward managing agents and strategic collaboration, with hiring weighing autonomy and judgment over specific technical expertise.

Hacker News front page

Show HN: 500 clocks in one 6.3 MB HTML file by Opus 5.5

开发者用 Claude Opus 5.5 在 Claude 聊天界面分多次会话生成了 500 个时钟,全部打包进一个 6.3 MB 的自包含 HTML 文件。作者称这是其个人版的「自行车上的鹈鹕」测试,用于检验前沿模型生成自包含网页的能力,并特别称赞 Opus 5.5 所造时钟的多样性与质量。

Hacker News front page

Show HN: Sphinx – A hook that quizzes you on your AI's code before shipping

Sphinx 是面向智能体辅助开发的免费开源 Git 钩子,在受保护分支推送前根据代码差异提问,答对后才继续推送。独立审查模型不读取编程聊天记录,回答不会传给编程智能体,也不会打开 .env 文件和密钥;可用 sphinx yolo 跳过检查,每次跳过都会记录。

AI HOT · Tips & opinions

Addy Osmani 谈 Claude Code 自验证:代码审查应转向多层验证

Addy Osmani 提出,当代码生成速度超过人工审阅能力时,代码审查应转向多层验证体系。他分享 Anthropic 团队经验,称提供验证手段可显著提升智能体产出质量,Claude Code 已能自己写 90% 以上的代码。验证体系包括 e2e 测试、property 测试和多智能体对抗式审查,人聚焦产品意图、风险与证据评估,并按风险分级审批。

TechCrunch · AI

Anthropic says its workforce has become less junior and is spending more time in meetings

Anthropic 周五发布的报告显示,其初级技术员工占比下降,员工人均会议时间较一年前增加 20%。自 2025 年初以来,技术部门初级员工占比从 21% 降至 14%,技术招聘中初级岗位占比从 27% 降至 9%。2026 年 4 月的调查中,27% 的员工表示 Claude 提高了自身生产力,20% 表示工作已从实际执行转向管理 Claude;报告指出,人类注意力与判断正成为生产力瓶颈。

Hacker News front page

tsc-ts: TypeScript written in TypeScript

tsc-ts 将 TypeScript 7.0.2 编译器(tsgo)移植为 TypeScript 实现,并通过 scriptc 编译为原生代码,可从 npm 安装后用 npx tsc-ts 运行。